[Pkg-utopia-maintainers] Bug#897607: dbus: Please raise timeout for a few test cases (or disable) in riscv64

[Simon McVittie]

On Sat, 05 May 2018 at 01:46:19 +0200, Manuel A. Fernandez Montecelo wrote:
> So the build failed in other tests now.

Based on these numbers I've tried a 20x multiplier for the timeouts on
__riscv in my recent upload to experimental. I've also added better
logging for buildd builds, so that I don't have to keep asking you
for logs.

If these changes are successful in experimental then I'll land them
in unstable.

I've assumed that this is a Debian-specific hack during bootstrapping,
and once bootstrapping is complete we'll either build on real hardware,
or cross-compile with nocheck and later test on real hardware? With my
upstream hat on, I don't think I particularly want to support running
the tests on a software emulation of a target CPU - if a test deadlocks,
I want it to fail in a reasonable time, and increasing the arbitrary
60 second timeout to 20 minutes doesn't seem like a very reasonable time.

> [many lines like these]
> dbus-daemon[4174]: [session uid=1000 pid=4174] The maximum number of
> pending replies for ":1.0" (uid=1000 pid=4094
> comm="/<<PKGBUILDDIR>>/debian/build-debug/") has been reached
> (max_replies_per_connection=128)

This is normal. It's one of several tests that hammers the dbus-daemon
to try to reproduce bugs.

    smcv